Indian Journal of Science and Technology

Abstract

Background/Objectives: Nowadays, many people enjoy the direct purchase on e-commerce which many kinds of item are increasing explosively. Existing method has a weakness of the accuracy to forecast, leaving customers unsatisfied. Methods/Statistical Analysis: We search association rule into customers’ buying behavior and discover items of group purchase which promoted cooperative buying. We can also create the table of association rules in the whole purchase history data to join customer’s record. If a customer would want additional sale for cross selling and up selling, the system could recommend the items which worked on the rate threshold of association rules to promote sale. Findings: We search association rule into customers' purchase data and discover frequent pattern to promote sale for selling associated items, to forecast frequent pattern of customer’s interest of item. We need to make clustering the category of associated items to reflect the customer’s propensity to reflect customer’s interest of associated item as well as to create the table of association rules in the whole purchased history data to join customer’s record. Our proposing system is higher 19.15% in F-measure than existing system. We execute the preprocessing for clustering the category of associated items to reflect the customer’s propensity. Our proposing system is higher 27.42% in recall, even if it is lower 12.23% in precision than previous system which conducted the analysis of segmentation to have different weights, is advanced than existing system. Our proposing system is better in F-measure than both the previous system and existing system. We could recommend associated items which worked on the rate threshold of association rules to promote sale, if a customer bought associated items. Improvements/Applications: We have the improvement that proposing system is higher 12.54% in F-measure than existing system. We make application to promote sale for selling associated items in e-commerce.
Keywords: Association Rules, Clustering, Collaborative Filtering, Segmentation Method